What your data department isn’t telling you… but your integration architect will

Central data governance is necessary, but it can’t scale fast enough for an AI-driven enterprise. Why the real key to semantic interoperability sits at the edge, with the systems and teams that create the data.

Published on
20 mei 2026
Written by
Koen Williame, Wim Paredis

The Illusion of Central Control

Ask your data department how to achieve semantic interoperability and you’ll likely hear about:

→ Building an enterprise-wide ontology

→ Defining canonical data models

→ Enforcing top-down vocabulary standards across the company.

Sounds noble. Strategic, even. And to be clear: it is needed.
Organizations do need common definitions. They do need governance. They do need alignment on what data means across teams, systems and business domains.

But in practice, it often feels like dead weight for delivery teams and application owners.

The truth is: central governance can’t scale fast enough to keep up with the speed and complexity of today’s digital ecosystem… By the time your data team finishes mapping your landscape, half of it has already changed.

And in an AI-driven enterprise, that gap becomes even more visible.

Because AI does not only need access to data. It needs access to meaning.

Meanwhile, your integration architect often advocates for a different approach. One that is lighter, faster, and, paradoxically, more sustainable.

Let’s break that down.

This is exactly where integration architects see the gap.

Semantic meaning cannot simply be added afterward. Once data is moving through the landscape, it becomes much harder to define, align, and govern consistently. The new imperative is to shift left: push ontologies, vocabularies, and semantic clarity as close as possible to the source.

In practice, that means encouraging suppliers, internal application teams, and SaaS providers to describe their data meaningfully from the start. The goal is no longer just to publish APIs. It is to make those APIs intelligible, self-describing, and semantically rich.

This bottom-up approach does not replace governance. It makes governance possible at scale.

Shift left: Where meaning should begin

When systems expose semantically rich interfaces, integration becomes easier, faster, and more scalable. Central teams can stop modeling and guessing, and start curating, aligning, and validating.

That is how governance evolves. Not as a function that invents meaning from the top down, but as one that consolidates and strengthens meaning already defined at the source.

In other words, the less your central team has to translate, the more room the organization creates for speed and innovation.

And that’s exactly the vision behind semantic interoperability: empowering every system to speak clearly so the organization can move forward as one.

A tale of two strategies

Let’s contrast the two approaches:

Build enterprise ontology centrally
Enforce standards across all teams
Require alignment before integration
Heavy upfront investment
Central bottleneck risk

Let systems describe themselves semantically
Provide standards and tools to teams
Allow integration with evolving semantic clarity
Progressive, scalable investment
Distributed ownership of meaning

Neither approach should exist in isolation.

Top-down governance brings consistency, control and enterprise-wide alignment. Bottom-up enablement brings speed, ownership and context.

The real opportunity is to combine both: central governance that sets direction, and distributed teams that define meaning where it originates.

That is where integration architecture plays a crucial role.

What needs to happen

To truly shift left, organizations need:

Standards like RDF, JSON-LD, and SHACL, but in a form development teams can actually use
API contracts that go beyond syntax and capture real meaning and intent
Tooling that makes semantic annotations part of normal development workflows, not an afterthought.
Architectural guidance that prioritizes composability, context, and clarity at the edge.

“
“Above all, this requires a mindset shift. Application teams should no longer be responsible only for exposing data, but also for making that data understandable.”
Koen Williame
Enterprise IT Architect

“
“And that mindset shift matters more than ever. Because in an AI-enabled organization, unclear meaning does not just slow down reporting or integration. It creates risk in the way AI systems interpret, recommend and act.”
Wim Paredis
Managing Partner

The future is emergent

Semantic interoperability is not a fixed state. It is a living capability. The strongest way to build it is not through control from the center alone. It starts with meaning defined at the edge, where data is created and understood.

So the next time your data department talks about a new governance initiative, ask your integration architect what they think.

Chances are, they are already working on the foundation for a more agile, scalable, and intelligent enterprise.

One where central governance does not carry the full burden alone.
One where systems describe themselves more clearly.
One where integration architecture connects local meaning to enterprise-wide understanding.
And one where AI can operate on more than just data.
It can operate on meaning.

What your data department is not telling you is that central control alone will not get you to interoperability. Your integration architect knows that real power sits at the edges, with the people closest to the data.

And in the age of AI, that edge is exactly where enterprise intelligence begins.